ALL:
Many flags in this topic and some posts removed. A reminder to keep this topic on-topic. If you want to continue participating, avoid posts aimed at other users or groups and focus solely on the discussion.
There’s plenty of bleating about it already. There’s even some evidence that it depends on which server you happen to be getting your weather from. All that bleating hasn’t budged anything anywhere though, so we sit here and discuss it amongst ourselves instead.
My only real issue with the current weather compared to 3 years ago is :
Weather transitions are not smooth anymore, clouds change shapes like ticking clocks one second at a time.
Instant sudden weather changes, from few clouds to suddenly a sky fully packed with dense clouds randomly mid flight.
Winds abruptly change as well alltho this is less now in current sim update.
We still have 1 single cloud type since release, just a cumulus cloud being stretched bigger or smaller.
Clouds appear to me more ash grey than they are white as per real life references. (I have a calibrated monitor with an average delta level of 0.2 with color accuracy below 0.5 average so definitely not me imagining things.
Storms have just been almost completely removed somehow from the sim, it’s extremely rare to see a lightning flash.
Percipitation aka rain doesn’t stay on the ground but simply dissapears in a few seconds after a shower.
The turbulent hot air rising from the big dense clouds can’t be felt in an airliner.
Sky and atmosphere is just the same no matter where in the world you are, zero pollution even tho meteoblue does provide this data.
Colorgrading on specific times near the golden hour are extremely red/pink on all altitude levels.
The list can go on for a while, this is from the top of my head at the moment.
It really surprised me seeing a small team from Laminar tackling all this while MS doesn’t in such a big time frame since release.
Weather is probably in the top 3 of most important aspects of a flight simulator, not just technically but visually as well.
Volumetric clouds can be done in multiple shapes and sizes, as proven by many other engines and DCS for example and soon XP in 12.07.
We’ve been bringing multiple issues up for over 2 years now nearing 3 about weather.
There were several improvements, but degredations that can’t be overlooked as well.
If anyone from the actual asobo dev team does somehow read this thread, i implore you… bring it up in a meeting and start prioritizing QOL updates that actually give meaning to a current gen flight simulator.
METAR weather is great, but u had something special with meteoblue and just completely trew that out of the window.
Vatsimmers can always use products like REX if they truly want METAR weather.
I agree that METAR broke a little bit live weather compared on how it was on release, but in the first releases of MSFS it happened very often to see weather COMPLETELY different in terms of QNH (pressure), wind, temperature and cloud coverage.
So something needed to be done.
I completely understand that METAR is not a perfect representation of weather also IRL, but trust me in the first releases of MSFS it happened so oftern that, for example, METAR reported wind 27008KT and on final you had 11010KT. Not acceptable.
But at release METAR was not part of data injected in the sim, so if MeteoBlue forecast was reporting 11010KT, having 11010KT in the sim was the expected result, it was not a bug. Knowing that, there was no reason to look for METAR information from an external source and try to compare it versus MeteoBlue.
Now that MeteoBlue integrates METAR, you really have risks of having different information (wind/speed/clouds/pressure, etc…) between those 2 sources, what creates bubbles. I don’t see how to avoid it having 2 different sources for the weather, what creates an unrealistic weather behavior. Even if I understand that request of having METAR, the experience I had with the initial Live Weather engine was more realistic than it is now.
The problem is the lag between reporting and injecting. Always has been. The reason that’s significant is explained in great detail above. The current solution is kludgy because all it does is highlight the lag, but it doesn’t fix the underlying problem.
To re-iterate:
A significant amount of users desire a way to use several methods of reported weather to “safely” navigate in real-time, as is done in the real world.
There are not enough tools in the sim to support this natively, so external, real-world reporting is often used.
The actual behavior of the real-world, reported weather will not match long-range forecasts to a degree which is realistic, so shorter-period injections, closer to short-range forecasts are desirable.
Injecting real-time METAR, often on top of and in front of the already injected weather highlights the lag. It also highlights areas of reported real-world weather that the MB injection is not picking up, for whatever reason.
Even more desirable, IMO, would be to get the injected lag down to 10-minute injections/corrections, which would eliminate the need for the METAR overlay and satisfy the vast majority of users because the weather would be behaving as it actually does, and would then be useable with real-world reports. The trick is then to animate/render it realistically.
Lastly, the cloud rendering and animation, lack of cirrus and CB (including lightning) is often conflated with this subject and deserves and should remain an entirely separate discussion. Animation and rendering is often confused with realism, despite non-realistic behavior. Make it animated, well rendered, and realistic. There’s no reason that can’t be the ultimate goal.
I didn’t vote for it either. What I voted for was plausible weather based on comparing in-sim weather with real-time observations so when the METAR says -10/-15 200 1/2 +SN you’re not looking at clear skies and 60 degrees when you load that airport up in the sim. That is what people were complaining about. In reality I can count on one hand the amount of time that the weather I came out in for ceilings and wind matched the ATIS, ASOS or METAR. But it would be “Oh, I broke out at 250 not 200” kind of difference - not “hm, the machine says it’s cloudy but there’s not a puff in 50NM of here” which is what I was experiencing. I firmly believe that something got lost in translation between what the user base was complaining about, and what they did to fix it.
Well, we should have compared the weather in sim with MB source. Not observations. And Asobo should have checked the data they got from MB to make sure it were injected propperly when we as users compared the weather with METAR. Because Asobo integrated MB source to begin with and knew for sure that were the source in use. It were not a bug that it didn’t matched METAR, it were a bug that it didn’t matched the MB source.
What you are pointing makes sense and maybe it is one or the root causes of actual Live Weather unrealistic behavior. Hopefully Asobo is conscious of that and is working on it.
The goal of an enhancement is to improve the user experience. METAR introduction was the opposite (from my point of view), it introduced a regression with those bubbles and also how clouds/layers were depicted even if we are not sure if it is related to the METAR but anyway it was part of the same regression. So I think Asobo knowing that (since they should go through quality testing cycles, testers should have noticed the regression) should have kept those changes in a BETA version until making sure to deliver a better user experience instead of a degraded one. Or if they were sure to fix the introduced regression within a couple of SUs (like 2 or 3), I have no issue to wait 6 months with a unrealistic Live weather engine in order to enjoy again the sim as I did at the beginning. But unfortunately there was no communication about that regression and after more that 1 year and a half we don’t have any sign what direction they are going to with Live weather, looks like it is not a priority, not so many users are complaining at the end. So the ones that feel the same as me are losing faith for short/medium term. Maybe at some point Live weather will be considered as an important feature again and we’ll see significant improvements but I see that at a long term.
We knew it didn’t match observations, and I’ve explained why that’s important to users, independent to a detailed explanation as to why that’s important to realism.
That it didn’t match the MB source goes back my long explanation that they were injecting the sim weather too far in advance of actual observed conditions, of which the specific outcomes could not be forecast at such a long range. This also explains why the weather would suddenly jump when it was updated - because it didn’t do exactly what the forecast predicted, or the forecast narrowed down in the ensuing time and a new forecast was issued based on that outcome. Either way, not a realistic weather experience.
I can see how that is the case. I agree it was kludgy and came back to highlight deeper, systemic issues. But it was also most likely the easiest temporary fix. Since that original fix it has gotten demonstrably better. The lag is much closer and the rendering is somewhat better. Continued incremental updates focusing on the areas we’ve discussed to death (summarized in my last post) could eliminate the need to inject METAR entirely.
I agree entirely. It is hard to explain all the things that go into weather and figure out how to resolve what was taking place in the sim (which we had little hard evidence, just observation) to how weather should and does behave in the world. So we got the kludgy METAR thing, which again, highlighted the fact they were using a larger forecast to generate specific weather. It wasn’t wrong to ask for better resolution, but my guess is like you said, people didn’t know what they didn’t know and how to ask for it.
Thankfully, it’s getting better. Let’s push for even better than that!
Is the sim capable of rendering complex layered weather? Yes. You can see that by creating custom weather.
Are the various platforms that use MSFS capable of rendering that complex weather? Yes. We do not see reports of users unable to run with complex weather (beyond the normal degredation).
Does the sim ingest the full range of weather data available? No. We know that MB only provide a segment of the available data.
Did weather match IRL souces well? No, in parts of the world with dynamic weather there were often significant discrepancies that could impact flight planning.
Did the sim provide native weather forecast or planning tools of comparable accessibility and design as basic IRL aviation weather products? No. And although these would be available via MB, this was not something obvious or readily accessible to users.
Was METAR well implemented? No - and this has been formally recognised, with attempts to fix the METAR bubble rendering.
Was it the implementation of METAR data sources that lead to the use of less complex cloud rendering? Perhaps. But I’ve seen no direct evidence to support that.
The continued availability of the rendering code for non-live weather and the absence of off-line weather performance issues suggest that the design limits to live weather were at root due to other factors than rendering or local platform performance. The game has been a phenomenal success and many of the SU5 and SU7 changes (not just weather, but eg TLOD also) suggest that server demand was far higher as a result requiring reductions in the calls on those servers to meet the vastly increased demand. This to me remains the most plausible explanation of the reduced fidelity of live weather. That, and some less than optimal implementation.
If it is server load management, then that is going to continue to have implications for the base performance of live weather and what may be achievable. But it also suggests that there could be other solutions - perhaps allowing the user to tailor their ‘live’ experience.
To me this is the real issue of design philosophy that should be questioned (as opposed to MB vs METAR). Why is it necessary for there to be a single ‘live’ experience with the consequent demand that creates for a single ‘live weather’ experience and the server demand that goes with it?
Some users will care more about fluid live weather based on a forecast model. Others will need ‘live’ weather that more closely matches IRL sources. Yet others will care more about other live players. Why is it necessary to make them all experience the same weather? Let them choose what kind of live experience they get through options, but at least recognize there is simply no need for a single, one-size fits all, lowest-common-denominator ‘live weather’.
I think you did a good summary of Live weather actual state, thank you. If the issue is a too high servers load and they decided to lower the data flow and sub-consequently the Live weather rendering then it would have been nice from Asobo to communicate us those reasons. The fact we have not received any explanation is just generating assumptions, speculations that lead at the end of doubting if they are planning to bring back the Live Weather to the level of realism, dynamism it was. I am not saying it was perfect, many points to improve but it was going toward the right direction in the 1st Sim Updates.
As you say, there is much speculation. I just personally found it doubtful that METAR implementation was the only cause of the changes we have seen. It may have aligned with other requirments though.
We need to move the debate beyond METAR vs MB, which has become sterile. I have no doubt that there were important reasons for the changes that were made and we continue to see incremental improvements. For a more fundamental change though, I think that calls for a re-think of what the ‘live’ environment should be. Which is more than just live weather. I made this thread to consider one possibility…
If i could suggest something that would increase the enjoyment for all of the users. Those that preffer the weather to stay as METAR says, Those who do not care, and those like me that preffer the fluidness of weather we had at release.
Option number 1:
12 hour weather model we had at release only that feels like a fluid. Would make it easier to report issues from MB source.
Option number 2:
METAR perfectly set around airports and should match the METAR 100% then a MB source around that no blending at all just pure METAR at airports and pure MB outside airports. Would make it easier to report issues from both types of sources.
Option number 3:
The hybrid we have now. Would stay equally hard to report issues as it is now.
I wonder if we have an issue with how this issue of degredation has been presented?
I cannot find an overarching ‘wishlist’ item for proposing changes to live weather. There are some very specific ones for tornadoes, improved colours etc. But nothing on the kinds of choices /options for live weather proposed at various points in this thread. But this thread is just a general discussion one - not a wishlist item.
There are other wishlist items for a weather API - but that has been clearly rejected.
There is also a dead-end wishlist item for a weather only update. I say dead-end as the mods have made it clear that the request makes no sense in terms of product development. Also it focuses on process and not what different features we actually want to see.
So I suggest there needs to be a new wishlist item opened, which is well framed and reflects the consensus view that the live weather should be:
predictable in line with IRL weather sources/forecast
Dynamic with smooth and organic development without bubbles or rapid transitions.
Make full use of the cloud rendering tools available in custom weather.
This is exactly what I’ve been trying to say. We’re focusing on the trees and missing the forest, getting stuck in minutia and placing blame on nonsensical or tangentially-related items, positing black or white options as if they are the only ones out there.
On the whole we’re missing two big-picture things: insight as to how the Asobo system really works under the hood (through no fault of our own), and insight into how weather (and its impact on aviation) really works, and folks are filling in the blanks to the limit of their personal ability. We should strive to have a better understanding of both and come up with a clear pathway instead of guesses and dead-end “fixes,” which is what got us to this point in the first place.
I was going to add that if all 3 are not achievable in one ‘live weather’ then users should be given options (as per @Perrry 's and my own posts above), for example: ‘high accuracy’ with a primacy on IRL sources or ‘dynamic’ with a focus on more seamless rendering, although really there should be no need for such a compromise: the real issue is the lack of cloud-type variety and layering at all altitudes.
[FWIW I personally find the live weather actually remarkablely accurate when compared with AWS forecasts for the PNW for ground wind speed & direction, temp, pressure, precip and vis and the winds aloft too. It’s the lack of cloud variety and layering that stands out. And proper TS /CB. The clouds just lack the dynamism and beauty of what was there pre-SU7]
Wishlist votes count - they make it to the top of the stack and get senior leadership focus which translates into actionn and development resource. But only if the request makes sense.
This is particularly important for weather, which is going to remain a core part of the SIM and with limits on any 3rd party alternative.
Well, there is only 2 types of weather data. Observed weather or forecasted weather. Before this sim we had either Observed only or observed together with forecasted weather. In this sim we had only forecasted weather to begin with. Right or wrong decission, i’m not sure but thats what we had. And i liked that approach because it doesn’t need to set the weather in an observed fixed state. Now thats gone and replaced with observed together with forecasted data and we have no option to set it back to forecasted only.
I know there is different methods to render the data they recieve and i agree that could be improved too, For example the low density of clouds. But increasing density of clouds will not make the data they render more varied.
Disagree completely . Did a flight last night over Philippines and flew through some amazing thunderstorms. The lightning bolts, the sound of thunder, the clouds lit up, the driving rain…all there.
